Lichess puzzles need some serious evaluation

Stockfish for puzzles! Thank you to the admins for this wonderful improvement.

To Frank200 :

Thrue, some problems of your list seem easy, but still I missed some of them : missed the fork on Problem 120056 / missed the solution of Problem 40605 / didn't see Qb5 mate on Problem 118665 / missed the fork on Problem 62258. So I am about 50% successfull on these problems, while my rating for tactics is around 1950.

My conclusion : some problems are easy for some people, tough for others. Depends on your weeknesses. I personally need to work on forks.


Re. 120056, is very useful for making these sorts of royal-fork setup motifs (and other motifs) second nature to you.

Re. 40605, it's not entirely trivial, but if you do enough tactics you should sense there may be a sacrificial breakthrough mate here; so many pieces aiming at the king is a red flag. On closer inspection, you'll see that f2 is holding white's position together -- so Rxf2+ is the tempo you need to check-check-checkmate on the g+h files. Your queen plus a single rook is enough for that (with the knight also controlling g3), so the rook is sacrificing itself on f2 for a worthy cause.

Re. 118665, the white king has ventured too far into your territory and is surrounded by squares controlled by your pawns (with e5 defended by your bishop). That should be a red flag to look for opportunities to further restrict and corral the opponent's king without letting him slip away (i.e., maintain control of squares around the king until you mate). Again, not entirely elementary but it's something you should keep in mind practically.

Re. 62258, the idea is a bit similar to 120056 in that there's not an immediate fork, but you give up some kind of exchange in order to create an immediate fork where you win back more material than you gave up. This is a practical concept in games as well as puzzles, so you should be on the lookout for opportunities to "give up" (not really) the exchange in order to create a winning fork.

In fact, all of these are practical ideas that can frequently be employed in games. Judging by your rating, you're familiar with all this, but perhaps you have to practice with these motifs more to see them quickly.

I want to thank Frank200 for such a lengthy answer. You've been very generous of your time to explain all this and very pertinent in your comments. It's nice to have players like you around.

If a puzzle has a high rating then many high rated players couldn't solve it. It is as easy as that. No need to make big lists.
Nobody ever complained: Oh I have a rating of 2000, but my real ability seems more like 1700 ... But player's ratings and puzzle's ratings are calculated the same way. The numbers theirselves may be incommensurable - a player with, say, blitz rating of 1700 might have puzzle rating of 2100. But i´there is always a reason why a puzzle or a player has a high rating.

The scoring overly penalizes incorrect answers. You seem to lose more for an incorrect solve than you get for a correct solve. Some recent statistics from me: 15 correct, 11 incorrect, lost 48 rating points; 7 correct, 4 incorrect, lost 1 rating point; 8 correct, 4 incorrect, lost 10 rating points.

True. It would be nice to put some info to explain all this. What is the statistical base for tactic problem solving rating calculation?

This topic has been archived and can no longer be replied to.